Release 303 (13 Feb 2019)#


  • Fix incorrect padding for CHAR values containing Unicode supplementary characters. Previously, such values would be incorrectly padded with too few spaces. (#195)

  • Fix an issue where a union of a table with a VALUES statement would execute on a single node, which could lead to out of memory errors. (#207)

  • Fix /v1/info to report started status after all plugins have been registered and initialized. (#213)

  • Improve performance of window functions by avoiding unnecessary data exchanges over the network. (#177)

  • Choose the distribution type for semi joins based on cost when the join_distribution_type session property is set to AUTOMATIC. (#160)

  • Expand grouped execution support to window functions, making it possible to execute them with less peak memory usage. (#169)

Web UI#

  • Add additional details to and improve rendering of live plan. (#182)


  • Add --progress option to show query progress in batch mode. (#34)

Hive connector#

  • Fix query failure when reading Parquet data with no columns selected. This affects queries such as SELECT count(*). (#203)

Mongo connector#

  • Fix failure for queries involving joins or aggregations on ObjectId type. (#215)

Base-JDBC connector library#

  • Allow customizing how query predicates are pushed down to the underlying database. (#109)

  • Allow customizing how values are written to the underlying database. (#109)


  • Remove deprecated methods getSchemaName and getTableName from the SchemaTablePrefix class. These were replaced by the getSchema and getTable methods. (#89)

  • Remove deprecated variants of methods listTables and listViews from the ConnectorMetadata class. (#89)